Faculty Bibliography

BACKGROUND:Internalizing and externalizing problems account for over 75% of the mental health burden in children and adolescents in the US, with higher burden among minority children. While complex interactions of multilevel factors are associated with these outcomes and may enable early identification of children in higher risk, prior research has been limited by data and application of traditional analysis methods. In this case example focused on Asian American children, we address the gap by applying data-driven statistical and machine learning methods to study clusters of mental health trajectories among children, investigate optimal predictions of children at high-risk cluster, and identify key early predictors. METHODS:Data from the US Early Childhood Longitudinal Study 2010-2011 were used. Multilevel information provided by children, families, teachers, schools, and care-providers were considered as predictors. Unsupervised machine learning algorithm was applied to identify groups of internalizing and externalizing problems trajectories. For prediction of high-risk group, ensemble algorithm, Superlearner, was implemented by combining several supervised machine learning algorithms. Performance of Superlearner and candidate algorithms, including logistic regression, was assessed using discrimination and calibration metrics via crossvalidation. Variable importance measures along with partial dependence plots were utilized to rank and visualize key predictors. FINDINGS/RESULTS:We found two clusters suggesting high- and low-risk groups for both externalizing and internalizing problems trajectories. While Superlearner had overall best discrimination performance, logistic regression had comparable performance for externalizing problems but worse for internalizing problems. Predictions from logistic regression were not well calibrated compared to those from Superlearner, however they were still better than few candidate algorithms. Important predictors identified were combination of test scores, child factors, teacher rated scores, and contextual factors, which showed non-linear associations with predicted probabilities. CONCLUSIONS:We demonstrated the application of data-driven analytical approach to predict mental health outcomes among Asian American children. Findings from the cluster analysis can inform critical age for early intervention, while prediction analysis has potential to inform intervention programing prioritization decisions. However, to better understand external validity, replicability, and value of machine learning in broader mental health research, more studies applying similar analytical approach is needed.

Background Angiotensin receptor neprilysin inhibitors (ARNI) reduce mortality and hospitalization for patients with heart failure. However, relatively high copayments for ARNI may contribute to suboptimal adherence, thus potentially limiting their benefits. Methods and Results We conducted a retrospective cohort study within a large, multi-site health system. We included patients with: ARNI prescription between November 20, 2020 and June 30, 2021; diagnosis of heart failure or left ventricular ejection fraction ≤40%; and available pharmacy or pharmacy benefit manager copayment data. The primary exposure was copayment, categorized as $0, $0.01 to $10, $10.01 to $100, and >$100. The primary outcome was prescription fill nonadherence, defined as the proportion of days covered <80% over 6 months. We assessed the association between copayment and nonadherence using multivariable logistic regression, and nonbinarized proportion of days covered using multivariable Poisson regression, adjusting for demographic, clinical, and neighborhood-level covariates. A total of 921 patients met inclusion criteria, with 192 (20.8%) having $0 copayment, 228 (24.8%) with $0.01 to $10 copayment, 206 (22.4%) with $10.01 to $100, and 295 (32.0%) with >$100. Patients with higher copayments had higher rates of nonadherence, ranging from 17.2% for $0 copayment to 34.2% for copayment >$100 (P<0.001). After multivariable adjustment, odds of nonadherence were significantly higher for copayment of $10.01 to $100 (odds ratio [OR], 1.93 [95% CI, 1.15-3.27], P=0.01) or >$100 (OR, 2.58 [95% CI, 1.63-4.18], P<0.001), as compared with $0 copayment. Similar associations were seen when assessing proportion of days covered as a proportion. Conclusions We found higher rates of not filling ARNI prescriptions among patients with higher copayments, which persisted after multivariable adjustment. Our findings support future studies to assess whether reducing copayments can increase adherence to ARNI and improve outcomes for heart failure.

BACKGROUND/UNASSIGNED:Nonadherence to antihypertensive medications remains a persistent problem that leads to preventable morbidity and mortality. Behavioral economic strategies represent a novel way to improve antihypertensive medication adherence, but remain largely untested especially in vulnerable populations which stand to benefit the most. The Behavioral Economics Trial To Enhance Regulation of Blood Pressure (BETTER-BP) was designed in this context, to test whether a digitally-enabled incentive lottery improves antihypertensive adherence and reduces systolic blood pressure (SBP). DESIGN/UNASSIGNED:BETTER-BP is a pragmatic randomized trial conducted within 3 safety-net clinics in New York City: Bellevue Hospital Center, Gouveneur Hospital Center, and NYU Family Health Centers - Park Slope. The trial will randomize 435 patients with poorly controlled hypertension and poor adherence (<80% days adherent) in a 2:1 ratio (intervention:control) to receive either an incentive lottery versus passive monitoring. The incentive lottery is delivered via short messaging service (SMS) text messages that are delivered based on (1) antihypertensive adherence tracked via a wireless electronic monitoring device, paired with (2) a probability of lottery winning with variable incentives and a regret component for nonadherence. The study intervention lasts for 6 months, and ambulatory systolic blood pressure (SBP) will be measured at both 6 and 12 months to evaluate immediate and durable lottery effects. CONCLUSIONS/UNASSIGNED:BETTER-BP will generate knowledge about whether an incentive lottery is effective in vulnerable populations to improve antihypertensive medication adherence. If successful, this could lead to the implementation of this novel strategy on a larger scale to improve outcomes.

BACKGROUND:Male sex, elevated troponin levels, and elevated D-dimer levels are associated with more complicated COVID-19 illness and greater mortality; however, while there are known sex differences in the prognostic value of troponin and D-dimer in other disease states, it is unknown whether they exist in the setting of COVID-19. OBJECTIVE:We assessed whether sex modified the relationship between troponin, D-dimer, and severe COVID-19 illness (defined as mechanical ventilation, ICU admission or transfer, discharge to hospice, or death). METHODS:We conducted a retrospective cohort study of patients hospitalized with COVID-19 at a large, academic health system. We used multivariable regression to assess associations between sex, troponin, D-dimer, and severe COVID-19 illness, adjusting for demographic, clinical, and laboratory covariates. To test whether sex modified the relationship between severe COVID-19 illness and troponin or D-dimer, models with interaction terms were utilized. RESULTS:Among 4,574 patients hospitalized with COVID-19, male sex was associated with higher levels of troponin and greater odds of severe COVID-19 illness, but lower levels of initial D-dimer when compared with female sex. While sex did not modify the relationship between troponin level and severe COVID-19 illness, peak D-dimer level was more strongly associated with severe COVID-19 illness in male patients compared to female patients (males: OR=2.91, 95%CI=2.63-2.34, p<0.001; females: OR=2.31, 95%CI=2.04-2.63, p<0.001; p-interaction=0.005). CONCLUSION/CONCLUSIONS:Sex did not modify the association between troponin level and severe COVID-19 illness, but did modify the association between peak D-dimer and severe COVID-19 illness, suggesting greater prognostic value for D-dimer in males with COVID-19.

Background/Purpose: Psoriatic arthritis (PsA) is a complex immune-mediated disease. Beyond its deleterious effects in the skin and joints, PsA can lead to decreased quality of life, increased psychosocial stress, and is associated with high levels of depression and anxiety. However, little is known about the effects of mental health on disease activity and severity. This may be especially important in PsA where up to half of patients have residual symptoms (i.e., pain, fatigue) despite effective immunomodulatory therapies. The objective of this study was to characterize the prevalence of psychiatric comorbidities and their impact on PsA outcomes in an urban, academic, combined clinic setting.
Method(s): Consecutive adult patients meeting CASPAR criteria (n=537) were prospectively recruited at the NYU Psoriatic Arthritis Center and followed for up to 2 years. All data was obtained from clinical visits using a standardized EPIC template. Depression was defined as patient-reported depression and/or use of anti-depressant medications.
Result(s): The cohort was 53% male, mostly Caucasian (79.7%) and had an average age of 49 years. Within our population, 23% had depression, 18% anxiety, and 4% ADHD (Table 1). At the initial visit, patients with depression were more likely to be female, older, and have concomitant anxiety compared to those without depression. Moreover, compared to their nondepressed counterparts, patients with depression had similar swollen joint counts (SJCs), tender joint counts (TJCs) and RAPID3 scores, as well as a lower percent body surface area (BSA). However, at the subsequent timepoints, while other outcomes remained similar between the groups, patients with depression had a higher TJC (Figure 1). When adjusting for age, sex, race, medication use, and comorbidities, the rate ratio (RR) of TJC in patients with depression vs. without depression was 1.23 (95%CI 0.78, 1.94, p=0.79) at baseline (Figure 2). This ratio was even higher at year 1 (RR 1.47, 95%CI 0.91, 2.35, p=0.19) and year 2 (RR 1.75, 95%CI 0.97, 3.14, p=0.07), nearing significance. In the adjusted models for SJC, BSA, and RAPID3, this pattern was not seen.
Conclusion(s): High rates of depression and anxiety in this cohort expand upon previously reported data. While most patients improve over time, TJC is significantly higher in those who carry a diagnosis of depression whereas SJC and BSA are similar in patients with and without depression. This may reflect differences in how patients with depression perceive their disease and may lead to difficulty in achieving low disease activity/remission by composite score measures. Therefore, addressing depression, along with inflammatory symptoms, should be considered, especially in those with residual pain. Further work is needed to understand if intervening on depression could help improve PsA outcomes

INTRODUCTION/BACKGROUND:The aim of this research was to assess the association between inflammation and oral health and diabetes, as well as the mediating role of oral hygiene practice in this association. METHODS:Data were from the 2009-2010 National Health and Nutrition Examination Survey. The analytical sample consisted of 2,191 respondents aged 50 and older. Poor oral health was clinically defined by significant tooth loss (STL) and periodontal disease (PD). Diabetes mellitus (DM) was determined by glycemic levels. The outcome variable was serum C-reactive protein (CRP) level, dichotomised as ≥1 mg/dL (elevated CRP) vs <1 mg/dL (not elevated CRP). Two path models, one using STL and DM as the independent variable, the other using PD and DM as the independent variable, were estimated to assess the direct effects of having poor oral health and DM on elevated CRP and the mediating effects of dental flossing. RESULTS:In path model 1, individuals having both STL and DM (adjusted odds ratio [AOR], 1.92; 95% confidence interval [CI], 1.30-2.82) or having STL alone (AOR, 2.30; 95% CI, 1.68-3.15) were more likely to have elevated CRP than those with neither STL nor DM; dental flossing (AOR, 0.92, 95% CI, 0.88-0.96) was associated with lower risk of elevated CRP. In path model 2, no significant association was found between having both PD and DM and elevated CRP; dental flossing (AOR, 0.91; 95% CI:, 0.86-0.94) was associated with lower risk of elevated CRP. CONCLUSIONS:Findings from this study highlight the importance of improving oral health and oral hygiene practice to mitigate inflammation. Further research is needed to assess the longer-term effects of reducing inflammation.

There is some, but inconsistent, evidence to suggest that matching patient treatment preference enhances treatment engagement and outcome. The current study examined differential preferences and factors associated with treatment preference for 12-week group cognitive behavioral therapy (CBT), yoga, or stress education in 226 adults with generalized anxiety disorder (GAD; 70% female, Mean age = 33 ± 13.5). In a subsample of 165 patients who reported an intervention preference and were randomized to yoga or CBT, we further examined whether match to preferred intervention improved the primary treatment outcome (responder status on Clinical Global Impressions Scale) and engagement (dropout, homework compliance). Preferences for CBT (44%) and yoga (40%) were similar among patients. Women tended to prefer yoga (OR = 2.75, p = .01) and CBT preference was associated with higher baseline perceived stress (OR = 0.92, p = .04) and self-consciousness meta-cognitions (OR = 0.90, p = .02). Among those not matched to their preference, treatment response was higher for those receiving CBT than yoga (OR = 11.73, p = .013); there were no group differences for those matched to their treatment preference. In yoga, those who received their preference were more likely to drop than those who did not (OR = 3.02, 95% CI = [1.20, 7.58], p = .037). This was not the case for CBT (OR = 0.37, 95% CI = [0.13, 1.03], p = .076). Preference match did not predict homework compliance. Overall, results suggest that treatment preference may be important to consider to optimize outcome and engagement; however, it may vary by treatment modality. Future research incorporating preference, especially with yoga for anxiety, is aligned with personalized medicine. TRIAL REGISTRATION: clinicaltrials.gov: NCT01912287; https://clinicaltrials.gov/ct2/show/NCT01912287.

For interventions that affect how individuals interact, social network data may aid in understanding the mechanisms through which an intervention is effective. Social networks may even be an intermediate outcome observed prior to end of the study. In fact, social networks may also mediate the effects of the intervention on the outcome of interest, and Sweet (2019) introduced a statistical model for social networks as mediators in network-level interventions. We build on their approach and introduce a new model in which the network is a mediator using a latent space approach. We investigate our model through a simulation study and a real-world analysis of teacher advice-seeking networks.